stunthanger.com
General control line discussion => Open Forum => Topic started by: Mark wood on January 12, 2022, 09:01:24 AM
-
I'm working on an electric powerplant control / timer and I'm fairly certain that a near Constant Velocity could be achieved. I'll publish that work eventually. It has been the focus of much of the testing I have been doing with the exception of the spade exercise which was a valuable distraction. In the process of the analysis, I make the assumption (simplification) of Constant Velocity, CV, in order to simplify the math involved. We always talk about that being desirable but I'm not so sure. While flying the SV videos I would swear the velocity is nearly constant but when I do the math and watch the video for verification, I see the airplane decelerating on the 45 lines. Which begs the constant Angular Rate question.
A Constant Angular Rate would make the squares look uniform and the leg time would be the same basically. With a CAR the upper level line would take the same time as the lower level line. With CV the top leg would be short time while the bottom leg would be long time. When we watch objects moving we don't necessarily see the actual speed so much as we see the characteristic length coved per unit time. A 767 on final looks like it is flying much slower than a Learjet 23 does even though both have nearly identical approach velocities. My truck going down the road at 80 MPH feels much slower than the wife's LaCross doing 80 MPH.
So, the question opened for discussion is CV or CAR as it relates to perception of the quality of the maneuvers.
-
What is the difference in length between the top leg and bottom leg of a square? What would be the airspeed difference between the two legs in order to get the times equal? How much slower would we need to fly the top leg and would the speed differential be great enough to cause stalling problems in the 3rd corner?
I’m assuming the CAR would apply to the vertical legs as well, accelerating (or increasing power) to keep the vertical up leg constant and slowing the down vertical leg accordingly. The CV mode should already attempt or address the time differential in the vertical legs. Do you think the CAR mode would be more effective than CV in the vertical?
-
When you say constant velocity, what is that relative to? The ground, or the air?
On squares, the top leg IS shorter than the bottom leg. Don't know how you could get to the same time by either method.
-
What is the difference in length between the top leg and bottom leg of a square? (Clip)
Though the rules for the AMA square loops do not specifically state that the top and bottom legs are to be the same length, the diagram does show both the top and bottom legs to be the same length. Since the rules do not require that the upright legs are to be perpendicular to the ground, the top two legs should be the same length given that the "Errors" section states that it is an error when the "Sides of loops are not equal". Higher scores should be given to those squares where all four legs appear to be the same length, given the other factors are the same (including size, smoothness, sharpness of corners, etc).
(There could be an interpretation where the Errors section states "Sides of loops are not equal" could mean only the upright sides. However, the upper side and the bottom side are also slides just like there is a left side and a right side. Perhaps the Errors section statement should be written "All sides of the loops are not equal". And here we are after just completing a three-year rules change cycle that had eleven change proposals, one would think the rules do not need more "adjustments".)
Keith
-
When you say constant velocity, what is that relative to? The ground, or the air?
On squares, the top leg IS shorter than the bottom leg. Don't know how you could get to the same time by either method.
Constant velocity of the aircraft. Airspeed. Since the square is a figure on a sphere, both top and bottom subtend the same angular change. For a constant angular rate the time it takes to transit that angle change would be the same. For CV the time on top would be less.
The question is semi hypothetical making an assumption of ability to achieve either.
-
What is the difference in length between the top leg and bottom leg of a square? What would be the airspeed difference between the two legs in order to get the times equal? How much slower would we need to fly the top leg and would the speed differential be great enough to cause stalling problems in the 3rd corner?
I’m assuming the CAR would apply to the vertical legs as well, accelerating (or increasing power) to keep the vertical up leg constant and slowing the down vertical leg accordingly. The CV mode should already attempt or address the time differential in the vertical legs. Do you think the CAR mode would be more effective than CV in the vertical?
Initially, the question posed would be for the level section, however, yes, assuming on all legs of the sphere. An underlying element of the question as posed is assuming a powerplant capable meeting the requirement of either DCV or CAR.
I am a powerplant engineer by trade. This is like an exercise in requirements capturing.
To answer your last question I don't exactly know which is why I ask the question. What I can say is that the vertical legs present some difficult solutions in terms of thrust increase and braking.
-
Though the rules for the AMA square loops do not specifically state that the top and bottom legs are to be the same length, the diagram does show both the top and bottom legs to be the same length. Since the rules do not require that the upright legs are to be perpendicular to the ground, the top two legs should be the same length given that the "Errors" section states that it is an error when the "Sides of loops are not equal". Higher scores should be given to those squares where all four legs appear to be the same length, given the other factors are the same (including size, smoothness, sharpness of corners, etc).
(There could be an interpretation where the Errors section states "Sides of loops are not equal" could mean only the upright sides. However, the upper side and the bottom side are also slides just like there is a left side and a right side. Perhaps the Errors section statement should be written "All sides of the loops are not equal". And here we are after just completing a three-year rules change cycle that had eleven change proposals, one would think the rules do not need more "adjustments".)
Keith
I presented the squares as an easy to comprehend way of thinking about the differences. There are a couple places in my videos where the deceleration can be seen but not necessarily perceived. One that is fairly evident is the approach to the outside squares. You might notice that the elevator control has a slight trend towards up elevator which is an indication of a loss of velocity. I don't yet have an airborne instrumentation package flying but I'm not far away from it. I've been using the videos for the understanding. Our club is going build several of Igor's indoor GeeBee plane, and I was watching his plane on this same approach and it had the same characteristic.
My intent is not to question the rules in any way. That is not what I set out for the discussion. It may drift that direction but I'm not going to drive it there. I'm primarily interested in the powerplant control aspect.
-
... Constant Velocity could be achieved. ...
... A Constant Angular Rate would ...
Be careful with your definitions.
I think you mean a constant rate of rotation as projected on the ground. That's one way of looking at things.
You probably don't mean the overall angular velocity -- that goes up in loops and especially in corners.
You probably also don't mean the angular velocity around the airplane's nominal yaw axis*, because if you do the math you'll find out that's exactly equal to the velocity in an earth-centered frame of reference, just as a consequence of geometry. Then if you actually think about how a plane flies, you'll find out that the actual angular velocity about the actual yaw axis isn't a good proxy for the nominal yaw axis velocity.
* By "nominal yaw axis" I really mean the axis that's perpendicular both to a line from pilot to plane and to the plane's velocity vector.
-
Be careful with your definitions.
I think you mean a constant rate of rotation as projected on the ground. That's one way of looking at things.
You probably don't mean the overall angular velocity -- that goes up in loops and especially in corners.
You probably also don't mean the angular velocity around the airplane's nominal yaw axis*, because if you do the math you'll find out that's exactly equal to the velocity in an earth-centered frame of reference, just as a consequence of geometry. Then if you actually think about how a plane flies, you'll find out that the actual angular velocity about the actual yaw axis isn't a good proxy for the nominal yaw axis velocity.
* By "nominal yaw axis" I really mean the axis that's perpendicular both to a line from pilot to plane and to the plane's velocity vector.
Definitely an engineer within this one. My thinking is more the flight path and not the aircraft axis'.
-
When you say constant velocity, what is that relative to? The ground, or the air?
On squares, the top leg IS shorter than the bottom leg. Don't know how you could get to the same time by either method.
Paul, as usual, beat me to it - with respect to ground/inertial space, or air. I would add - I would not count on any accelerometer you can fit in a stunt plane to be adequate to integrate the acceleration over an entire 6.5 minute stunt flight. So if you really mean WRT inertial space, you are going to need some absolute velocity sensor to measure it. I can think of two likely candidates, but one measures the airspeed and the other measures the groundspeed.
For the record, my own design (hypothetical only) tries to control the inertial velocity using the Y axis accelerometer as a measure of the inertial velocity for relatively slow corrections to the integral of the acceleration, then uses a PI controller to drive the motor (which is presumed to be a 1/s plant).
Brett
-
Paul, as usual, beat me to it - with respect to ground/inertial space, or air. I would add - I would not count on any accelerometer you can fit in a stunt plane to be adequate to integrate the acceleration over an entire 6.5 minute stunt flight. So if you really mean WRT inertial space, you are going to need some absolute velocity sensor to measure it. I can think of two likely candidates, but one measures the airspeed and the other measures the groundspeed.
For the record, my own design (hypothetical only) tries to control the inertial velocity using the X axis accelerometer as a measure of the inertial velocity for relatively slow corrections to the integral of the acceleration, then uses a PI controller to drive the motor (which is presumed to be a 1/s plant).
Brett
So what about the observer perspective? This input is what you're doing doing and closing loop on acceleration. Most all of the approaches if not all of the approaches to the motor control take on the architecture of an autopilot system running an auto throttle. The filtering necessary to make an IMU work and the powerplant response create a fairly large lag between the onset of speed droop and thrust correction leading to overshoot and oscillation. Given that a corner of a square is not very long, any thrust lag is significant.
So, the question is really what is it we desire? If we could create a control that could fly at a desired mode.
-
So, the question is really what is it we desire? If we could create a control that could fly at a desired mode.
I could not tell you the search terms for that, but if you do some digging you may find some discussion about that headed up by Igor Burger.
The Burger timer is not the first iteration of his attempts at electronic control of speed regulation for electric stunt -- he describes going through some iterations including, but not limited to, controlling the motor like a piped engine (i.e. using a speed control with an unstable zero in the disturbance loop, so that the motor would go faster under load), yaw rotation rate, and, finally, y-axis (outward from center of circle) acceleration.
He settled on y-axis acceleration because that worked best for him.
If you can find it, read it -- it's a very good map of the potential potholes on the road to joy for any would-be engine regulator designer (like -- me, if I ever get back to it).
Personally, having read what Igor has said, and users of Fioretti timers, and hearing rumors about some of the top US folks, I'm thinking that a speed control needs to widely tunable, and given that the trick will be to find a way to tune it so that you don't need to be a control systems engineer to make the plane behave the way you want.
Also personally, I'm very curious of what it would be like to fly a plane that's servoed very close to constant speed in inertial (ground-referenced) space. I know that if you hand something like that to an experienced pilot they're unhappy because the pacing of the maneuvers is just all wrong. But I'm wondering if you deliberately go there and then practice up to it if you can't do a better pattern after the (probably) year or two of acclimating yourself to the plane's behavior.
-
what is it we desire?
What I desire is a power plant that actually decelerates when heading down hill but maintains (but does not accelerate) speed going uphill.
-
I was just wondering if Mark is aware of the résumés of the participants in this conversation, not to mention the cumulative man-years they've spent just on square-maneuver-corner thrust profile.
-
I have been working with the Igor control system for 9 years now.
I first saw it at the 2012 world champs. I was amazed by the amount and rate of the changes in motor speed on his and several others. My first thought was that I was not sure it helped the corners. I purchased a system from Igor and was off experimenting with all the options. And options abound! I fell into the same trap everyone else who has had one does. That being I wanted to hear and feel it do its thing by speeding up and slowing down. I clearly won two Nat's because of the system, and feel I also gave away one, clearly, and likely a second one also. It is easy to go overboard, as is very addictive. For the longest time I have wondered why mine does not function like Igor's. My conclusion is that it comes down to the trim of the plane. Igor's yaws noticeably, and his system, as has been pointed out already, responds to Y axis accelerations. His plane yawing about thus responds more than my plane which I trim to give very steady line tension. I have tried every imaginable combination to make it work, including his "pipe" mode.
Saying all that, I have found that using too much brake is a sure fire way for me to bounce corners. The system just does not respond fast enough to get back on the power to keep the corner flat. My solution has been to reduce the sensitivity, but not eliminating it, to "keep" the power on to keep the corners flat. Even with the low sensitivity, it doesn't run away in the wind coming downhill. My sensitivity is enough to know it is there, but not enough to hear. It is my feeling that this is optimal for my style of trim and piloting.
As far a a square maneuver being the same length on all four sides, use the engineers favorite check by taking things to the limit. A square maneuver by definition has vertical sides that are perpendicular to the ground. Now, consider a 90 degree tall square loop. It is 90 wide. Follow those sides vertically to 90 degrees, and what do you get? They intersect, forming a three sided square. Oops..three sides? If the "square" is limited to 45 degrees, the top horizontal leg IS shorter than the bottom level flight leg. If you use timing, to do your square, it will not be correct. Gravity against you going up, shorter leg on the top, gravity with you going down, and the bottom leg longer than the top. All contribute so one shouldn't use your internal metronome for this maneuver.
-
Saying all that, I have found that using too much brake is a sure fire way for me to bounce corners. The system just does not respond fast enough to get back on the power to keep the corner flat. My solution has been to reduce the sensitivity, but not eliminating it, to "keep" the power on to keep the corners flat. Even with the low sensitivity, it doesn't run away in the wind coming downhill. My sensitivity is enough to know it is there, but not enough to hear. It is my feeling that this is optimal for my style of trim and piloting.
For reference, I found the same issue with incorrect IC engine setup, if it backs off too much coming down the hill, it almost guarantees that it will hop out of the corners. Before I lost the bubble completely (as noted the weekend before Golden State) I was adjusting that with oil content, more oil, less "backoff". I think the mechanism is pretty clear, it suddenly backs off the power, reduces the control loading, which causes you to go harder over with the controls than you were expecting, hop. I think you need it to maintain some reasonably consistent control pressure.
BTW, the same thing was what struck me about my first electric flight with Bobby's Genesis, the control loading was positive, and consistent, in every corner, instead of some of them being soft and just "falling" into the corner, and having to put in positive force to get it to go. This was with no feedback at all. just the RPM regulator - so no backoff at all.
Of course, at the NATS and our pre-Golden State practice session, I had the opposite problem, it was charging in the corners, making it nearly impossible sometimes. Had I left that alone and then tried to fly it in the 25 MPH wind, I would certainly have stuck it in the ground - I was barely making it at 15. Switched setups, back to good again for the real contest.
Brett
-
Paul, I am curious on your opinions regarding the governor-only type timers (Hubin FM-9/Castle esc) when compared directly against the lowered sensitivity parameters you have used with the Igor system.
As a thought exercise, compare as if these systems were used in same plane, same conditions. What/where/how do you feel the governor-only system would be lacking. Which maneuvers would suffer or gain the most? Ect.? What/where/how would the Igor type system be easily superior? You have a distinct perspective from your experience testing and competing to comment on the differences.
-
So what about the observer perspective? This input is what you're doing doing and closing loop on acceleration. Most all of the approaches if not all of the approaches to the motor control take on the architecture of an autopilot system running an auto throttle. The filtering necessary to make an IMU work and the powerplant response create a fairly large lag between the onset of speed droop and thrust correction leading to overshoot and oscillation. Given that a corner of a square is not very long, any thrust lag is significant.
So, are you suggesting/proposing some sort of feedforward (presumably from the bellcrank or Z axis of the accelerometer), to get more lead?
the problem that seems to be common with all of these systems is that you are still dependent on the motor controller/throttle bandwidth, which sets a upper limit on the acceptable feedback control. You appear to be suggesting a feedforward of some sort to torque the motor directly rather than send a signal to a separate controller and allowing it to adjust the RPM. That is at least similar to my (completely notional) idea, where the controller *is* the accelerometer, there is no separate RPM-regulating loop that you are adjusting up and down.
So, the question is really what is it we desire? If we could create a control that could fly at a desired mode.
That is the critical question, as Paul notes, you might not necessarily want a perfect controller, and using the nonlinearities to adjust the ideal response might make a perfect controller a moot question.
Brett
-
... I fell into the same trap everyone else who has had one does. That being I wanted to hear and feel it do its thing by speeding up and slowing down. ...
Soooo, an Igor timer plus a speaker playing engine noises, that sound more labored when the 'lectric motor is being asked for more power?
It could be done!
edit:
I think you could even do that with all-analog electronics, or mostly-analog. That'd be -- interesting.
-
The problem that seems to be common with all of these systems is that you are still dependent on the motor controller/throttle bandwidth, which sets a upper limit on the acceptable feedback control. ...
A continuing frustration for me is that I don't have time to design an ESC for this. I've got the control systems and motor drive chops, there's open-source designs out there that just* need software. I think that -- possibly at the expense of needing a custom tuning for each motor/prop combination, or an adaptive controller -- you could drive the bandwidth way higher than what ESCs do now.
Or (and this isn't my original idea) a constant-speed ESC and a variable-pitch prop driven by a fast servo.
* It's never "just software" -- but it's so easy to say!
-
What I desire is a power plant that actually decelerates when heading down hill but maintains (but does not accelerate) speed going uphill.
That is doable. A variable pitch propeller is necessary to achieve it.
-
I was just wondering if Mark is aware of the résumés of the participants in this conversation, not to mention the cumulative man-years they've spent just on square-maneuver-corner thrust profile.
I am.
-
A continuing frustration for me is that I don't have time to design an ESC for this. I've got the control systems and motor drive chops, there's open-source designs out there that just* need software. I think that -- possibly at the expense of needing a custom tuning for each motor/prop combination, or an adaptive controller -- you could drive the bandwidth way higher than what ESCs do now.
Or (and this isn't my original idea) a constant-speed ESC and a variable-pitch prop driven by a fast servo.
* It's never "just software" -- but it's so easy to say!
I'm not certain a fast servo is even required.
-
I'm not certain a fast servo is even required.
Overall, I don't either.
But if you want to keep up with the plane's loss of velocity in a corner -- yes, it is.
-
That is doable. A variable pitch propeller is necessary to achieve it.
Yes, and that is a much more direct way of controlling the thrust than changing the RPM. In that case the prop inertia helps you rather than hurts you. Igor tells us that the prop inertia is a driving problem for the conventional controller design, presumably because of finite torque/finite current or current slew rate. I think that came up in the thread about folding props, it came up in the "electronic controls" rule discussion, and several similar SSW threads.
The actual numbers matter (which I don't have) but I would expect that the pitch control servo could achieve much higher Fdot (rate of change of the thrust) than trying to spin the prop up or down.
Brett
-
So, are you suggesting/proposing some sort of feedforward (presumably from the bellcrank or Z axis of the accelerometer), to get more lead?
the problem that seems to be common with all of these systems is that you are still dependent on the motor controller/throttle bandwidth, which sets a upper limit on the acceptable feedback control. You appear to be suggesting a feedforward of some sort to torque the motor directly rather than send a signal to a separate controller and allowing it to adjust the RPM. That is at least similar to my (completely notional) idea, where the controller *is* the accelerometer, there is no separate RPM-regulating loop that you are adjusting up and down.
That is the critical question, as Paul notes, you might not necessarily want a perfect controller, and using the nonlinearities to adjust the ideal response might make a perfect controller a moot question.
Brett
Since Howard brought up resumes, I feel compelled to share a bit of mine pertinent to the conversation. I am aware of the level of talent here.
I am after all a powerplant engineer and I've been involved with Full Authority Digital Engine Controls, FADEC, since Allison built their first lunch box computer version in 93. I've worked on two tiltrotors, several helicopters, a bunch of jets, some power stations, a boundary later blower for a flying boat, a twin engine pack single propeller and some off the wall things. Shaft engines from 50 Hp to 4,500 Hp and thrust engine from 400 pounds to 26,000 pounds thrust. I have a little insight. I'm not and likely never will be the stick I am currently in the midst of. But I can make a better powerplant. That I am ultimately confident in. One of those projects is the RAH66 Comanche helicopter. It required the chase vehicle to be a Corvette in order to keep up with its side wards acceleration. That takes some doing from the powerplant control system perspective and it can't accomplished using a droop based control loop. That was one of the vehicles I had a hand in.
Bare with me. I have not yet reached a complete analysis in order to present here on what I am working on. This is something I started last year and it included running a variable pitch propeller which is why I did the hinged propeller testing which precipitated that precession discussion. I'll refrain from using constant speed because that is not what is necessary for constant CV AR or other given criteria. Think of variable pitch as a way of shifting the efficiency around for a given velocity and power requirement. We'll trek that road another day.
When Howard posted his data, there were two elements that drew my attention one of them was the thrust response which can be seen in the x axis data. More importantly the lag and subsequent following rate error. The very first part of the chart the x accel is showing a bit of a negative level, likely from some yaw. Shortly the pitch rate ramps up from the control input and is seen in the x axis deceleration. I made a couple thin black lines to show the offset and the droop. Granted this is acceleration but the speed surely followed as the drag set in. At a point the x axis droop turned around and increased. It's hard to know exactly what is transpiring and it would be good to have motor command in this plot. I drew a green line to represent what I think is going on in terms of thrust response. I have a couple Fiorotti timers and have worked with them enough to know they aren't much different. This lag is inherent in the filtering required to get a low noise signal from the IMU whether it is rolling average or a Kalman filter or some other method.
I'm a powerplant controls guy. I'm gonna state this. The droop governor is for steady state operation. This chart is an example of why we have other control loops for transient operation. Maneuvering is not steady state.
So, let me pose this question. What would happen if we could match the thrust output to the thrust required to overcome the drag of maneuvering? What if it could be done without over shooting and minimal phase lag?
Part of the difficulty of trimming the power output of the G or rate sensing systems I'm seeing is that he gain is linear. At least the Fiorroti seams to be when I play with it on the bench connected to my Oscope. I'm not certain how much I'm going to continue to mess with it beyond flying some with the data logger.
Drag is not linear with G. So, when the gain is g = kG there will be times when there is overthrust or under thrust for a given condition. This is impossible to trim without having lots of velocity variation. My active timer is turned way down. Given the phase lag of the error development and the thrust errors this system won't ever meet the dynamic demands perfectly during maneuvering.
-
Yes, and that is a much more direct way of controlling the thrust than changing the RPM. In that case the prop inertia helps you rather than hurts you. Igor tells us that the prop inertia is a driving problem for the conventional controller design, presumably because of finite torque/finite current or current slew rate. I think that came up in the thread about folding props, it came up in the "electronic controls" rule discussion, and several similar SSW threads.
The actual numbers matter (which I don't have) but I would expect that the pitch control servo could achieve much higher Fdot (rate of change of the thrust) than trying to spin the prop up or down.
Brett
Yeah I had some conversation with Igor on that. Apparently the active timer has better performance with higher spin motors due to the lower resistance. My propeller testing has been on hold as the test airplane was involved in a catastrophic landing. A controllable pitch propeller would be the cat's meow in terms of performance which has been my journey for the last year and some change. That is, in fact, the end game of what I'm working on.
-
Since Howard brought up resumes, I feel compelled to share a bit of mine pertinent to the conversation. I am aware of the level of talent here.
I wasn't the one discussing resumes and I wasn't challenging your assumptions, just trying to find out what you were talking about and maybe offering some suggestions - so I am not sure why you are responding to my post. As I mentioned the other day, conveying information inherently demands asking questions, and those questions are not necessarily challenges or "assertions in the form of a question". I am very happy to discuss it with anyone with an idea - if it is good idea, it doesn't matter where it came from. In any case, I am sure Howard's CV looks better than mine; if nothing else his would be mostly unclassified.
I noted that you could use feed-forward to reduce/eliminate the phase shift inherent in the closed-loop control. You can make it a function of anything you want. My notion from a few years ago would be to use the Z axis acceleration/load factor as either an additional lead element, or, as a feedforward, since it more-or-less measures the drag for you. You would need a band-pass filter to block DC and high frequency noise, but otherwise you could make the feedfoward signal proportional to the absolute value of load factor, the square of the load factor or whatever else you think it should be.
But your first question is still the most critical - what is it that we are trying to control? Paul's observation, I think, relates to excessive/excessively aggressive/excessively nonlinear/out of phase control of the airspeed, where the pitch sensitivity changes rapidly. I think it is the same issue from the "what does 'penetration' mean?" thread. So I might be inclined to start on the premise you are trying to control the body frame X acceleration, that is, the "orbit rate", for lack of a better name for it, and using the Z acceleration as a feedforward and maybe the Y acceleration as a lead element in some sort of a lead-lag control system.
If you are going to control the "orbit rate" which is probably what you want with the "constant angular rate" idea, you *do* have to have some other system to sense it directly, hence, Y acceleration (which at low frequency is proportional to the integral of the X acceleration due to our constraint).
But that is just notional, I am not sure if it is well thought through. The original idea was to use the X acceleration to replicate a Fox 35, which is kind of how this all started.
Brett
p.s. this (and most similar) threads point out an even more basic engineering issue - there are tremendous gaps translating the engineering to the subjective handling qualities that most of us think are the goal. Put another way, we have never defined "good" VS "bad" in an engineering sense, even to first approximation. Paul's example (and my similar example) illustrate that in droves. Paul knew exactly what he changed, and I had at least some idea, but why that improved the cornering performance is at best educated guessing. In Paul's case, very highly educated, but nonetheless, a guess). This is a problem with full-scale aircraft, too. This is the basis of the Cooper-Harper Handling Qualities ratings - a subjective scale used to determine how difficult an airplane is to fly. Of course, the pilot has to be qualified to provide these ratings, and while Paul is probably the best possible evaluator, any of the more experienced/accomplished stunt pilots probably give reasonable evaluations.
But it *is* subjective. I have flown two flights 10 minutes apart, one where the airplane would have rated a 2 (good, negligible deficiencies), and another where it maybe rated a 5 (moderately objectionable). The differences - same airplane, same engine, same conditions, absolutely no trim changes, just altering the oil content of the fuel by 4%. Do the same the next weekend in different conditions, and I might get the same ratings - for exactly the opposite change!
Another example, a 9 (Major Deficiencies, intense pilot compensation required to retain control) to about a 2. Again, the same airplane one minor trim change, and one minor engine setup difference. Several people were witness to me nearly crashing my airplane at the bottom of a triangle in 15 mph wind, same airplane with 1/2 ounce of weight on the tail and a slight pipe length change, it handled 25 with not much problem.
Point being, while I definitely try to figure out what causes these differences, we don't really have any way to translate that into engineering parameters, so it is very difficult to do more than guess at the right answer to your original question, and then try it and see. Electric helps because there is at least some hope of knowing what changed in reasonable engineering terms, why it mattered is still subject to speculation.
-
I wasn't the one discussing resumes and I wasn't challenging your assumptions, just trying to find out what you were talking about and maybe offering some suggestions - so I am not sure why you are responding to my post. As I mentioned the other day, conveying information inherently demands asking questions, and those questions are not necessarily challenges or "assertions in the form of a question".
I noted that you could use feed-forward to reduce/eliminate the phase shift inherent in the close-loop control. You can make it a function of anything you want. My notion from a few years ago would be to use the Z axis acceleration/load factor as either an additional lead element, or, as a feedforward, since it more-or-less measures the drag for you. You would need a band-pass filter to block DC and high frequency noise, but otherwise you could make the feedfoward signal proportional to the absolute value of load factor, the square of the load factor or whatever else you think it should be.
But your first question is still the most critical - what is it that we are trying to control? Paul's observation, I think, relates to excessive/excessively aggressive/excessively nonlinear/out of phase control of the airspeed, where the pitch sensitivity changes rapidly. I think it is the same issue from the "what does 'penetration' mean?" thread. So I might be inclined to start on the premise you are trying to control the body frame X acceleration, that is, the "orbit rate", for lack of a better name for it, and using the Z acceleration as a feedforward and maybe the Y acceleration as a lead element in some sort of a lead-lag control system.
But that is just notional, I am not sure if it is well thought through. The original idea was to use the X acceleration to replicate a Fox 35, which is kind of how this all started.
Brett
That wasn't exactly intended for you, sorry. And I don't feel attacked or otherwise by you Brett. You guys are very talented individuals. However when that is blatantly thrown in to my face I feel compelled to respond in kind. That it appeared to be directed towards you, wasn't meant to be. Just a knee jerk reaction on my part.
My original question is subjective and is about what is it we are really after. A constant velocity, for instance, would cause some interesting visual appearances and the model would look to be hauling a$$ up high. Which brings the alternate question, would a constant angular velocity be better. Both the upper and lower sides of the square have the same angular displacement. So, flying at constant angular velocity or close would maybe have a better appearance. That is my question. It wasn't intended to go down the road of controlling a motor. My bad for stating that I came up with this question during an analysis I am performing.
As is what I have stated, I began this with a notion of what would it take to truly make an airplane fly constant velocity and consequently I worked up the basic needs in terms of physical hardware. A motor controller and variable pitch propeller. I'm not convinced that attitude is necessary for good performance. Given the correct flexibility of the powerplant without a constant velocity or angular velocity constraint requirement I'm 85% confident it isn't.
For instance take my question. If the powerplant can provide the correct thrust to compensate for increased drag during maneuvers then the velocity variation becomes simply a function of the change in gravitational potential. ie Delta V from level flight would be the square root of two times the change in height. That can be improved some fairly simply I think and I haven't yet finished working through that portion. Which will need some inertial computations to achieve. This is why I asked the question the way I did was to be able to decide whether to pursue that aspect or not.
I've actually made some comments in the past about the approach because it is derived from how we use anticipation and acceleration power control in the FADECs. I think that the discussion on the actual control and gain strategy should be a separate thread.
-
That wasn't exactly intended for you, sorry. And I don't feel attacked or otherwise by you Brett. You guys are very talented individuals. However when that is blatantly thrown in to my face I feel compelled to respond in kind.
As noted the other day, I think you are reading far more into this than was intended. Noting other people's capabilities does not mean that he was questioning yours.
In any case, whether it was intentional or just perceived, getting wound up about it doesn't help you, me, or anyone else advance the state of the art. You have interesting observations and ideas, don't derail your point on side issues .
Brett
-
As noted the other day, I think you are reading far more into this than was intended. Noting other people's capabilities does not mean that he was questioning yours.
In any case, whether it was intentional or just perceived, getting wound up about it doesn't help you, me, or anyone else advance the state of the art. You have interesting observations and ideas, don't derail your point on side issues .
Brett
I will do my best. Apologies Howard.
I posted an article in a separate thread walking through the derivation of an open loop gain control. Bottom line that drag is a fairly simple equation 1-Cos X which lends itself readily to a simple gain. An input using a pot or hall effect sensor to pick off control position is all that is necessary to set thrust. We use this type of thing in the turbine engines when we loose a sensor or we're providing some acceleration anticipation for load. The loop closes naturally with the pilot in the loop. No need for an outside measurement. This initial article is without any inertial input. In order to improve the performance getting a G vector is necessary simply to know if the airplane is upright or inverted and turning upward or downward. In that case a trimming gain would be added to the primary gain. I don't believe that is necessary because this should get the airplane through to corners with most of it's original energy depending on the response of the motor. Clearly using this approach running a propeller has huge potential. The real question is whether that is necessary or not. Having flown the Fiorotti now for a couple hundred flights, my jury is out on continuing the variable pitch effort. My current effort is building the instrumentation package followed by finishing the airplane design. This timer control was on hold for a while but Howards data provoked me to progress it further at this time since that data gave me a piece of information I was looking for.
-
I posted an article in a separate thread walking through the derivation of an open loop gain control. Bottom line that drag is a fairly simple equation 1-Cos X which lends itself readily to a simple gain. An input using a pot or hall effect sensor to pick off control position is all that is necessary to set thrust.
1-cos(x), where the 1 is a normalized bias representing the parasitic drag and the cos(x) is the induced drag, all multiplied by V^2 somewhere else? In any case, I think you want to filter out the DC components, so maybe you don't need the bias/parasitic drag.
So, the answer to my earlier question is that you are using the control position as the feed-forward signal, the theory being that if the controls deflect, this will inevitably lead to the generation of induced drag, slowing the airplane - open-loop compensation (on top of anything else), around any closed-loop control.
It is a very interesting idea, and there is some experience with that sort of system from the mists of time - Scott Bair's modified ST46 with an exhaust throttle actuated by bellcrank position. This was entirely mechanical, if the bellcrank deflected either way, it opened the exhaust throttle. This worked more-or-less OK compared to any of the similar systems using a carburetor for the throttle, because the exhaust throttle seemed to work much faster.
Using it for electric is likely to work a lot better, for the same reason - less lag. The limiting factor is likely to be frying the output transistors in the motor controller and/or running out fo current, because you probably want to run your feedforward right to the base of the driver transistor.
My alternate idea was to do something similar using the Z acceleration as a direct measure of the load factor/induced drag.
In either case, make sure you can adjust the transfer function from the bellcrank to the delta-current easily with software changes because I think, as noted above, that "perfect" compensation might not be the best practical solution - which is again a restatement/the implication of Paul's comment. More than just the gain parameter, the entire transfer function (in case it wants to be linear, or a square root function, or something else, when you test it).
Excellent thought process, and definitely food for thought and grounds for further investigation.
Brett
-
. My propeller testing has been on hold as the test airplane was involved in a catastrophic landing.
Yeah, I hate when that happens!
A controllable pitch propeller would be the cat's meow in terms of performance which has been my journey for the last year and some change. That is, in fact, the end game of what I'm working on.
This sounds like a great idea. I would like to make a point here. If you have tried one of Igor's hollow props, you know how light they are. I commented to Igor how much I liked them because they reduced the gyroscopic precession significantly. His response was that he didn't make them for that reason, but to reduce the inertia the motor had to overcome to speed up and slow down. For me, it was one of the quantum changes that improved my flying and the planes performance. In building a variable pitch propellor, I am concerned that the gyro effects will get worse, not better. Not sure how you could build that arrangement for 16 grams. A higher mass prop in the corners concerns me because of the yaw. Something to consider.
-
1-cos(x), where the 1 is a normalized bias representing the parasitic drag and the cos(x) is the induced drag, all multiplied by V^2 somewhere else? In any case, I think you want to filter out the DC components, so maybe you don't need the bias/parasitic drag.
So, the answer to my earlier question is that you are using the control position as the feed-forward signal, the theory being that if the controls deflect, this will inevitably lead to the generation of induced drag, slowing the airplane - open-loop compensation (on top of anything else), around any closed-loop control.
It is a very interesting idea, and there is some experience with that sort of system from the mists of time - Scott Bair's modified ST46 with an exhaust throttle actuated by bellcrank position. This was entirely mechanical, if the bellcrank deflected either way, it opened the exhaust throttle. This worked more-or-less OK compared to any of the similar systems using a carburetor for the throttle, because the exhaust throttle seemed to work much faster.
Using it for electric is likely to work a lot better, for the same reason - less lag. The limiting factor is likely to be frying the output transistors in the motor controller and/or running out fo current, because you probably want to run your feedforward right to the base of the driver transistor.
My alternate idea was to do something similar using the Z acceleration as a direct measure of the load factor/induced drag.
In either case, make sure you can adjust the transfer function from the bellcrank to the delta-current easily with software changes because I think, as noted above, that "perfect" compensation might not be the best practical solution - which is again a restatement/the implication of Paul's comment. More than just the gain parameter, the entire transfer function (in case it wants to be linear, or a square root function, or something else, when you test it).
Excellent thought process, and definitely food for thought and grounds for further investigation.
Brett
It's all software based and a basic Arduino nano can host it. The actual control gain looks like gain = K1*(1-K2*Cos (k3 * theta)). Pass that through a dead band Gain out = abs(gain) >Kfilt will filter most of the noise and any that gets through isn't much of a problem since the powerplant has inertia. The output from the control / timer is a simple PWM to the ESC so there is some filtering there too. While it is open loop to the motor, the actual loop is control input, gain, motor, airplane, pilot. If the airplane is slow the elevator will be deflected inputting a command for power causing the airplane to accelerate. Think of the control deflection being the error signal generator.
I posted my incomplete evaluation:
https://stunthanger.com/smf/engineering-board/electric-motor-control-severe-geek-speak/msg629304/#msg629304
-
My propeller testing has been on hold as the test airplane was involved in a catastrophic landing.
Yeah, I hate when that happens!
A controllable pitch propeller would be the cat's meow in terms of performance which has been my journey for the last year and some change. That is, in fact, the end game of what I'm working on.
This sounds like a great idea. I would like to make a point here. If you have tried one of Igor's hollow props, you know how light they are. I commented to Igor how much I liked them because they reduced the gyroscopic precession significantly. His response was that he didn't make them for that reason, but to reduce the inertia the motor had to overcome to speed up and slow down. For me, it was one of the quantum changes that improved my flying and the planes performance. In building a variable pitch propellor, I am concerned that the gyro effects will get worse, not better. Not sure how you could build that arrangement for 16 grams. A higher mass prop in the corners concerns me because of the yaw. Something to consider.
I am sure even you can't build it for 16 grams. But as noted above, with the variable pitch setup, a higher propellor inertia actually helps you as far as controlling the speed goes, because it will tend to mitigate the prop slowing down/speeding up as you change the pitch. Whether it is better enough speed-control wise to make up for the extra precession, only one way to find out.
Can't help you directly on the precession - however, an earlier thread also notes that in addition to controlling the overall pitch, you could *hypothetically* vary the pitch up and down through 1 rev - just like collective and cyclic control on a helicopter, and use that to generate both pitch torque in the direction you want to go, and also, provide a yaw torque to compensate for the precession.
Effectively you would be using the cyclic control to reorient the prop angular momentum vector instead of forcing it around with the rest of the airplane. To first approximation, it wouldn't matter how heavy/how much moment of inertia the prop might have - which is fortunate, because the swashplate and other mechanics would make it even heavier. There are several very obvious flaws with this idea, not the least of which is trying to pitch the blades up and down at 400 Hz and this pounding the motor mount, swashplate, swashplate follower, prop hub, etc, to dust.
Either way sounds like a lot of pretty precise machining, which more-or-less lets me out of acting on any of my grand ideas.
Brett
-
In any case, I am sure Howard's CV looks better than mine; if nothing else his would be mostly unclassified.
Not even close. Brett is a national stunt champ and—um—a responsible person.
My notion from a few years ago would be to use the Z axis acceleration/load factor as either an additional lead element, or, as a feedforward, since it more-or-less measures the drag for you. You would need a band-pass filter to block DC and high frequency noise, but otherwise you could make the feedfoward signal proportional to the absolute value of load factor, the square of the load factor or whatever else you think it should be.
The square: synthetic induced drag. You could find out experimentally what the sign should be. It might be a function of wind.
-
Well, except when you put a hinge at the root of the propeller, then the precession is nearly eliminated. A pinned joint can create no moment. A pinned and hinged blade creates no moment and therefore doesn't precess. I made a brash statement of such without a good demonstration and created a real sit storm. Probably deserved some of it. I have been on a very long road towards this power system. So, I was worried about fatigue from bending of the blade from the continuous precession of the propeller during flight with the CL model. Each blade undergoes a bending cycle every revolution during flight and it doesn't take long to get a few million bending fatigue cycles. Most applications don't undergo this much continuous load.
If the blade spindle is a small bolt it must carry that bending. So, the idea came to me that one of my F1C folding propellers had a hinge and that would eliminate the bending. I did a test to see if the propeller could follow sharp corners. When I did, I realized there was little precession evident. I asked a question, in the wrong tone, about how significant is the precession for the CLPA airplanes, meaning is it something worth spending time and effort on. Well... Flight testing there is a noticeable improvement for some of the planes I tried it on. While others not so much. The low TVCv airplanes really tell you about it. My future plan is to use the Fuse cam and fly the prop on another airplane. I'm also working on the datalogger which will have an IMU, throttle control system load cell and position for ubergeekdom data generation. I don't have a version of the prop yet created for the large airplanes. I'm waiting on some more observation flights before I make that commitment.
Here's a link to my discussion on the hinged propeller:
https://stunthanger.com/smf/stunt-design/propeller-precession/msg620424/#msg620424
-
This sounds like a great idea. I would like to make a point here. If you have tried one of Igor's hollow props, you know how light they are. I commented to Igor how much I liked them because they reduced the gyroscopic precession significantly. His response was that he didn't make them for that reason, but to reduce the inertia the motor had to overcome to speed up and slow down. For me, it was one of the quantum changes that improved my flying and the planes performance. In building a variable pitch propellor, I am concerned that the gyro effects will get worse, not better. Not sure how you could build that arrangement for 16 grams. A higher mass prop in the corners concerns me because of the yaw. Something to consider.
I am sure even you can't build it for 16 grams. But as noted above, with the variable pitch setup, a higher propellor inertia actually helps you as far as controlling the speed goes, because it will tend to mitigate the prop slowing down/speeding up as you change the pitch. Whether it is better enough speed-control wise to make up for the extra precession, only one way to find out.
Can't help you directly on the precession - however, an earlier thread also notes that in addition to controlling the overall pitch, you could *hypothetically* vary the pitch up and down through 1 rev - just like collective and cyclic control on a helicopter, and use that to generate both pitch torque in the direction you want to go, and also, provide a yaw torque to compensate for the precession.
Effectively you would be using the cyclic control to reorient the prop angular momentum vector instead of forcing it around with the rest of the airplane. To first approximation, it wouldn't matter how heavy/how much moment of inertia the prop might have - which is fortunate, because the swashplate and other mechanics would make it even heavier. There are several very obvious flaws with this idea, not the least of which is trying to pitch the blades up and down at 400 Hz and this pounding the motor mount, swashplate, swashplate follower, prop hub, etc, to dust.
Either way sounds like a lot of pretty precise machining, which more-or-less lets me out of acting on any of my grand ideas.
Brett
I'm not sure of the camming portion of this you imagine. Placing a hinge in the blade root eliminates 90%+ of the precession. Much lighter than a swash plate. I actually have a 10x6 version that has been flying. I had two before the incident. Fatigue is a huge concern when it builds up at 400 hz.
-
Using it for electric is likely to work a lot better, for the same reason - less lag....My alternate idea was to do something similar using the Z acceleration as a direct measure of the load factor/induced drag.
Wouldn't one of these tend to goose the throttle at the bottoms of round loops and the other at the tops?
-
Any resume would have extensive gaps and blank records for bonuses, promotions, etc. All my records from 1983 to 2002 are completely blank - because they were in my paper file, and they didn't bother to transfer it for the electronic system. The others have a bunch of dollar values, awards, and nondescript "promotions" for no adequately explained reasons. I just got semi-roped into another stretch - but the end is in sight.
Of course, this all supports my jet-set lifestyle, I am an International Hollywood Playboy, just like Jethro on the Hillbillies.
"Will perform conformal mapping for food."
-
Wouldn't one of these tend to goose the throttle at the bottoms of round loops and the other at the tops?
The low-pass filter/estimator would have to be carefully defined. In this hypothesized system you are just using it to correct for the drift you get from integrating the X acceleration. Despite Marks earlier comments, the very last thing I would use is a variable-gain filter, and in practice, it's running at a constant frequency with all the input continuously available, the gains would rapidly wind up at steady-state values anyway.
Note that this is a strictly hypothetical idealized situation - my expectation is that by the time you put in enough limiters and saturations to keep it out of trouble, you are going to be riding one or the other of them almost continuously. You don't necessarily even want perfectly controlled speed no matter which speed you decide to control, as Paul notes.
I like the idea of the feedforward, you might just use the existing governor system, feed Mark's signal forward, and use the same signal into the governor to keeo them from fighting each other. So, more-or-less, Igor's system with an additional feedforward signal.
Brett
-
The low-pass filter/estimator would have to be carefully defined. In this hypothesized system you are just using it to correct for the drift you get from integrating the X acceleration. Despite Marks earlier comments, the very last thing I would use is a variable-gain filter, and in practice, it's running at a constant frequency with all the input continuously available, the gains would rapidly wind up at steady-state values anyway.
Note that this is a strictly hypothetical idealized situation - my expectation is that by the time you put in enough limiters and saturations to keep it out of trouble, you are going to be riding one or the other of them almost continuously. You don't necessarily even want perfectly controlled speed no matter which speed you decide to control, as Paul notes.
I like the idea of the feedforward, you might just use the existing governor system, feed Mark's signal forward, and use the same signal into the governor to keeo them from fighting each other. So, more-or-less, Igor's system with an additional feedforward signal.
Brett
That's a true statement. Probably the biggest trouble is the liner gain with G. The gain function I wrote would just as appropriately work with other systems. Simply inserting the 1-Cos into the gain loop would make a significant improvement in performance. The phase shift would still be there though.
In the turbine engines we use the same kind of strategy with the exception that the gain is derived from a look up table or a function several points along the curve. Look at my other thread and the CL v AOA op line is essentially how we drive the inlet guide vanes. In those controls we compute multiple gains continuously include this particular anticipation gain and they all are being feed to a highest or lowest wins gate. So in normal operation the control is operating on the "droop governor" but when a large control input is made the "anticipation gain" wins the gate and takes over. The ESC governors actually work quite well and the output of the timer control actually works as a reset for that governor.
-
Well, except when you put a hinge at the root of the propeller, then the precession is nearly eliminated. A pinned joint can create no moment. A pinned and hinged blade creates no moment and therefore doesn't precess. I made a brash statement of such without a good demonstration and created a real sit storm. Probably deserved some of it. I have been on a very long road towards this power system. So, I was worried about fatigue from bending of the blade from the continuous precession of the propeller during flight with the CL model. Each blade undergoes a bending cycle every revolution during flight and it doesn't take long to get a few million bending fatigue cycles. Most applications don't undergo this much continuous load.
If the blade spindle is a small bolt it must carry that bending. So, the idea came to me that one of my F1C folding propellers had a hinge and that would eliminate the bending. I did a test to see if the propeller could follow sharp corners. When I did, I realized there was little precession evident. I asked a question, in the wrong tone, about how significant is the precession for the CLPA airplanes, meaning is it something worth spending time and effort on. Well... Flight testing there is a noticeable improvement for some of the planes I tried it on. While others not so much. The low TVCv airplanes really tell you about it. My future plan is to use the Fuse cam and fly the prop on another airplane. I'm also working on the datalogger which will have an IMU, throttle control system load cell and position for ubergeekdom data generation. I don't have a version of the prop yet created for the large airplanes. I'm waiting on some more observation flights before I make that commitment.
Here's a link to my discussion on the hinged propeller:
https://stunthanger.com/smf/stunt-design/propeller-precession/msg620424/#msg620424
Sounds interesting. I will be watching for you flight test results.
-
So, more-or-less, Igor's system with an additional feedforward signal.
Maybe Igor's system canceled during a corner. Igor adds throttle with +X acceleration.
-
This sounds like a great idea. I would like to make a point here. If you have tried one of Igor's hollow props, you know how light they are. I commented to Igor how much I liked them because they reduced the gyroscopic precession significantly. His response was that he didn't make them for that reason, but to reduce the inertia the motor had to overcome to speed up and slow down. For me, it was one of the quantum changes that improved my flying and the planes performance. In building a variable pitch propellor, I am concerned that the gyro effects will get worse, not better. Not sure how you could build that arrangement for 16 grams. A higher mass prop in the corners concerns me because of the yaw. Something to consider.
Suppose you could reduce the mass of the rotating machinery by a factor of (I'm just pulling an integer out of the air) 4. That would reduce moment of inertia by 16. 4 of those little masses might be equivalent in power to the original rotating machinery, but with 1/4 of the moment of inertia, even if they rotate the same way.
-
One thing to keep in mind, and some of you do, is that some stuff varies with the square of airspeed and some stuff remains constant with airspeed. I imagine a doodad hooked up to the flap control horn with a bobweight-actuated slider to work the throttle. Details left as an exercise.
-
Suppose you could reduce the mass of the rotating machinery by a factor of (I'm just pulling an integer out of the air) 4. That would reduce moment of inertia by 16. 4 of those little masses might be equivalent in power to the original rotating machinery, but with 1/4 of the moment of inertia, even if they rotate the same way.
Now why would you be mentioning this??🤣🤣🤣🤫🤫🤫🤫
-
Now why would you be mentioning this??🤣🤣🤣🤫🤫🤫🤫
I think because it would be entertaining. If I wasn't already doing several other projects, I might would do it just for the entertainment value.
-
I think because it would be entertaining. If I wasn't already doing several other projects, I might would do it just for the entertainment value.
I believe that Howard got his stick out, sharpened it, and poked me with it based on what we talked about last week.
-
I believe that Howard got his stick out, sharpened it, and poked me with it based on what we talked about last week.
Ah. I was back one post and totally missed it. I was thinking about the bobweight doodad which I have some flyweight governor work in my history. Before FADECs there were mechanical systems which are very interesting. That input almost provoked me in to setting my brain on to working out the solution.
-
One thing to keep in mind, and some of you do, is that some stuff varies with the square of airspeed and some stuff remains constant with airspeed. I imagine a doodad hooked up to the flap control horn with a bobweight-actuated slider to work the throttle. Details left as an exercise.
Interestingly so does G.
-
I am following the ongoing discussion on automatic speed controllers with great interest and have great respect for the quality of the arguments put forward by the experts. It is this kind of objective discussion which maintains and promotes our common cause. Many thanks to the participants and to Stunthangar for maintaining the website.
According to what has been published here so far, it could be that after a complex development process we will see e a rather complex and not cheap system. I therefore take the liberty of proposing a first step to gain knowledge first:
A prerequisite for further steps should be that in order to design an automatic speed control system, the term speed should be defined. Whether the purpose of the controller should be to keep constant a given value for the airspeed, the angular speed or the groundspeed remains to be determined.
In order to support the opinion making on which speed to aim for, I believe that it would be useful to allow experiments on as broad a possible basis. This could be done by an adjustment of the making legal manual in-flight speed control by the pilot. For electric drives (and possibly IC motors, too) the current technology available would make it possible to make such systems available to a wide range of pilots at low cost.
I am positive that wide spread experience gained from manual in-flight speed control would be helpful defining requirements for the development of automatic speed controllers.
-
I am following the ongoing discussion on automatic speed controllers with great interest and have great respect for the quality of the arguments put forward by the experts. It is this kind of objective discussion which maintains and promotes our common cause. Many thanks to the participants and to Stunthangar for maintaining the website.
According to what has been published here so far, it could be that after a complex development process we will see e a rather complex and not cheap system. I therefore take the liberty of proposing a first step to gain knowledge first:
A prerequisite for further steps should be that in order to design an automatic speed control system, the term speed should be defined. Whether the purpose of the controller should be to keep constant a given value for the airspeed, the angular speed or the groundspeed remains to be determined.
In order to support the opinion making on which speed to aim for, I believe that it would be useful to allow experiments on as broad a possible basis. This could be done by an adjustment of the making legal manual in-flight speed control by the pilot. For electric drives (and possibly IC motors, too) the current technology available would make it possible to make such systems available to a wide range of pilots at low cost.
I am positive that wide spread experience gained from manual in-flight speed control would be helpful defining requirements for the development of automatic speed controllers.
Actually Peter what I am doing can be accomplished with an Arduino based board. There's zero automation. Basically the output can be tuned to match what the airplane is required to do. I published my thoughts quite a bit prematurely and this precipitated something much more complex than is necessary. My original purpose of asking the question was to gain insight on what I need to present in terms of performance prediction. As it is, a simple input from a potentiometer can be read and used to create the command. I use gain as the control would have a base level output for level flight. Then any needs of the airplane would commanded by the 1 - Cos X gain which, as can be seen, is a function of drag and can be made to fit quite well. This means that the control can output the thrust as required for the airplane to not loose energy from the turning. This gain could be adapted to a feedback controller but my experience as a powerplant engineer tells me it's not necessary for what we are doing. This same architecture is what wee use in helicopter engine control as failure mode accommodation which works nearly as well or as good as the hydro mechanical control systems.
What this means is that, if the system is trimmed correctly, the exit of the turns will only decay in velocity by the change in gravitational potential energy. I did not present this quite yet as this is the portion of my article I am working on. I am working through the mathematic derivation and the illustrations to demonstrate. Basically, once we can match the drag, the analysis becomes similar to a frictionless change in total energy. The question becomes one of were does the work come from which is obviously the powerplant. It is extremely hard to describe in words and requires the illustrations and mathematic derivation. If you haven't read my thread on motor Control, please have a read there.
My initial Energy conversion yielded some 28 ish fps velocity delta. The only computation I have worked through is simple, frictionless change in altitude where the change in kinetic energy from the change in potential energy drives the velocity change. Offhand that seems like allot but I have no real intuitive feel for it initially. I know the airplane slows down allot from watching the videos now, but I never really perceived that deceleration until I went looking for it and it verifiable in my videos. This is what begets the thread title question. The following use 65' base circle.
For instance, in level flight the airplane is going 78 fps which translate to about 69 Deg/s (Dps). The delta V of 28 fps means the airplane would be flying at about 50 fps on the 45 degree azimuth. This then translate to 289/360 = 0.8, and 50 fps / 0.8 D/ft = 62.5 Dps. Our perception of velocity is a function of the characteristic length of the thing we're watching and perceive the airplane as flying at a near constant velocity. So what I think we like is that the airplane crosses the high leg at near the same angular rate. I.e. we'd like to see i flying at a near constant 69 Dps in this case. I'm not so sure and that is the foundation of my question, because the small change I've derived is consistent with my observation, although not extensive.
In the simple version of my control, there is no need for any attitude information. However, in order to compensate for the mgh term in the energy a means of doing work to mitigate the loss is necessary. Basically we need to add or decrease thrust as a function of the pitch attitude as a function of the angle between the airplane and the gravity vector. This again doesn't need to be a complicated computation other than filtering the input from the IMU.
Take a square non tethered loop as an example. When we are in level flight we make the assumption Thrust equals Drag, T=D. In order to transit the corner we add thrust to compensate drag and maintain T=D. That compensation is the 1-cos X term. If that term is exactly correct, it won't be, then the exit would be slower by the change in altitude 1/2mV^2 = mgh leads to dV =sqrt(2*g*dh). For the non tether case during the turn the g vector changes as Sin X. It's not so simple in 3 space. In order to mitigate the gh term we simply need to add a trimming function to the gain computation Kg*Sin X so that the power gain would now be K1*(1-K2*Cos(K3*X)) + Kg*Sin X. With this we can now fine trim the airplanes velocity to suit what our perception wishes.
The need to know the g vector is only necessary to make the small improvement. In upset training I teach my student the importance of drag in the recovery and I demonstrate the difference between pulling 2 G's from vertically nose down and 4 G's nose down. In the first case we increase velocity tremendously while in the second case we actually slow down.
I my perspective, the impact of the IMU is mostly detrimental. This is due to two things. First is the amount of filtering required to get a good signal which creates a lag in response. Second is the encroachment in the processor timing margin which also leads to increased lag. Actually those two combined are the source of the lag. What I am finding on the Arduino products is that the smaller ones won't load and run the calibration libraries. I'm a total amateur when it comes to code and electronic hardware and have always relied on the others in my team for that.
Thanks for brining this up and letting me have the floor to present more of this effort. After I manage to update my article with this portion, I'll present the end game concept which incudes the variable pitch propeller.
-
According to what has been published here so far, it could be that after a complex development process we will see e a rather complex and not cheap system.
I think the hardware will not be terribly expensive, not much different from Igors or the others. The firmware will take some development, but once you have it, reproducing it is free. So I don't recurring costs will be anything unreasonable - there will certainly be less actual production cost than, say, an AAC 61 with a pipe and a carbon prop. It's almost a miracle that it only costs $500.
I also very strongly expect that when it comes down to it, the control system itself will be vastly simpler than some of the ideas displayed here - just following the real-life experience, most users have wound up restricting the range of adjustments and features to the point that the feedback only slightly alters the system from a fixed RPM governor. Same development curve as piped engines - yes, you can make it do stuff, but overdoing it is *extremely common*. I have seen a few of these "better" systems that were wildly unstable limited only by something saturating and limit-cycling.
I think of threads like this more like a thought experiment, but you can't design it from scratch like we had requirements to meet, because we have never defined any requirements and only have vague notions of why some settings work better than others.
The one thing that seems to be clear (since Igor pointed it out 5 years ago and Paul and others have confirmed it) that the current system is limited by the ability to spin the prop up or down. An obvious development step is to do something to reduce or remove that restriction and then see what effect it has. That seems to be what Paul and Howard are working on.
Brett
-
I think the hardware will not be terribly expensive, not much different from Igors or the others. The firmware will take some development, but once you have it, reproducing it is free. So I don't recurring costs will be anything unreasonable - there will certainly be less actual production cost than, say, an AAC 61 with a pipe and a carbon prop.
I also very strongly expect that when it comes down to it, the control system itself will be vastly simpler than some of the ideas displayed here - just following the real-life experience, most users have wound up restricting the range of adjustments and features to the point that the feedback only slightly alters the system from a fixed RPM governor. Same development curve as piped engines - yes, you can make it do stuff, but overdoing it is *extremely common*.
I think of threads like this more like a thought experiment, but you can't design it from scratch like we had requirements to meet, because we have never defined any requirements and only have vague notions of why some settings work better than others.
The one thing that seems to be clear (since Igor pointed it out 5 years ago and Paul and others have confirmed it) that the current system is limited by the ability to spin the prop up or down. An obvious development step is to do something to remove that restriction and then see what effect it has. That seems to be what Paul and Howard are working on.
Brett
Possibly, but this will work on a $7 Arduino running C+. From my point of view, this is what I am working on. If it gets taken to a different place that's great. The real test is when it flies and that is a project for after I finish coding the datalogger. I actually have an operational Arduino timer now with the radio interface. It isn't a big change to get one to fly. The hardest code branch is the gain tuning input. My intent is to initially fly without an IMU. Reducing the propeller mass is always good and I'd sure like to see how they are doing the mold process.
-
Possibly, but this will work on a $7 Arduino running C+.
Ugh, C++, the bane of my existence, and the source of the holes burned through my stomach lining? There is no utility or value to object-oriented code in an embedded processor, and if it is not object-oriented, you don't need C++ (vice C or other procedural languages).
But yes, that was my point, the hardware is not going to be expensive.
Brett
-
Ugh, C++, the bane of my existence, and the source of the holes burned through my stomach lining? There is no utility or value to object-oriented code in an embedded processor, and if it is not object-oriented, you don't need C++ (vice C or other procedural languages).
If you want to have that discussion, we should take it outside.
Folks -- even software professionals who should know better -- confuse C++ done wrong with C++ done right. Like Fortran done wrong, or C done wrong, or Ada done wrong, or machine code done wrong, or any engineering process whatsoever that's done wrong, C++ done wrong is just crap. C++ done right is every bit as good as those other things done right, but "done right" means, in large part, understanding what the @#$% you're doing.
And if it's done right it can let you write safe and elegant code for larger systems than you can in C (no comment on Ada or modern Fortran, because I've only debugged Ada code* and have only written Fortran for school assignments, back in the '80's).
If you're seeing bad C++, you need to fire the software engineers that are doing it wrong and hire a crew that knows how to write code for embedded systems and are willing to work to understand how compilers work -- just changing languages without swapping out the components in the process that are causing the real trouble won't fix things.
Almost every new line of code in the airborne FLIR products coming out of Portland was written in C++ starting in around 1985 1995 (edit: got my dates wrong). By 1990 2000 or so it was nearly every line of code in the product lines coming out of FLIR Portland. The code base was far bigger than it was when we were writing in C, the bug rate was about the same, the man hours per line of code was about the same, but the lines of code per bit of functionality was way higher -- meaning that management got way more bang for the buck.
Because we** made sure that when we adopted it, we did it right.
* Written by someone with an opinion about C similar to yours about C++, and too damned arrogant to look at her own code.
** "We" meaning me and a fellow engineer who had been writing object-based*** code for years without knowing it, and who actually know how the tools work, and who care enough to know how embedded code has to work.
*** Not to be confused with "pure" object-oriented code -- that's bad in a deeply embedded system.
-
If you want to have that discussion, we should take it outside.
Folks -- even software professionals who should know better -- confuse C++ done wrong with C++ done right. Like Fortran done wrong, or C done wrong, or Ada done wrong, or machine code done wrong, or any engineering process whatsoever that's done wrong, C++ done wrong is just crap. C++ done right is every bit as good as those other things done right, but "done right" means, in large part, understanding what the @#$% you're doing.
And if it's done right it can let you write safe and elegant code for larger systems than you can in C (no comment on Ada or modern Fortran, because I've only debugged Ada code* and have only written Fortran for school assignments, back in the '80's).
If you're seeing bad C++, you need to fire the software engineers that are doing it wrong and hire a crew that knows how to write code for embedded systems and are willing to work to understand how compilers work -- just changing languages without swapping out the components in the process that are causing the real trouble won't fix things.
Almost every new line of code in the airborne FLIR products coming out of Portland was written in C++ starting in around 1985 1995 (edit: got my dates wrong). By 1990 2000 or so it was nearly every line of code in the product lines coming out of FLIR Portland. The code base was far bigger than it was when we were writing in C, the bug rate was about the same, the man hours per line of code was about the same, but the lines of code per bit of functionality was way higher -- meaning that management got way more bang for the buck.
Because we** made sure that when we adopted it, we did it right.
* Written by someone with an opinion about C similar to yours about C++, and too damned arrogant to look at her own code.
** "We" meaning me and a fellow engineer who had been writing object-based*** code for years without knowing it, and who actually know how the tools work, and who care enough to know how embedded code has to work.
*** Not to be confused with "pure" object-oriented code -- that's bad in a deeply embedded system.
Yeah, but I'm not the code guy. I can hack out a functional control using Arduino IDE which is the top of my ability. In fact I already have one. For an amateur user, such as myself, I can build what I need. I have no need for Level A, B or C software for this application. If you're willing to help me on that level, I would very much appreciate it but I'm on this road with focus and the next step is flying the control open loop on control input.
-
Yeah, but I'm not the code guy. I can hack out a functional control using Arduino IDE which is the top of my ability. In fact I already have one. For an amateur user, such as myself, I can build what I need. I have no need for Level A, B or C software for this application. If you're willing to help me on that level, I would very much appreciate it but I'm on this road with focus and the next step is flying the control open loop on control input.
Not critiquing your choice -- first make it work, then make it work good. The nice thing about CL is that assuming you're plane is in good condition, the worst that a software error will do is splat a pretty plane into concrete.
Just getting tired of Brett's rant, which appears to be based on some bad experiences and apparently without detailed knowledge of what he's critiquing. I've ran into it before -- in the 1990's. It's interesting to see it arising now.
-
Not critiquing your choice -- first make it work, then make it work good. The nice thing about CL is that assuming you're plane is in good condition, the worst that a software error will do is splat a pretty plane into concrete.
Just getting tired of Brett's rant, which appears to be based on some bad experiences and apparently without detailed knowledge of what he's critiquing. I've ran into it before -- in the 1990's. It's interesting to see it arising now.
Thanks for that. There's some of this that I'm stretching on and code is where that is. Elliptical integrals is the other.
-
Just getting tired of Brett's rant, which appears to be based on some bad experiences and apparently without detailed knowledge of what he's critiquing. I've ran into it before -- in the 1990's. It's interesting to see it arising now.
"Without detailed knowledge of what I am critiquing"? Probably right, aside from the code review/"self-induced bug" hunting I was in 2 hours ago.
Of course you can write proper embedded processor code in C++ - by forgetting the words "abstraction", "object" and just about everything that has ever been touted as an "advantage" of C++. Superset of C, right?
Brett
-
Thanks for that. There's some of this that I'm stretching on and code is where that is. Elliptical integrals is the other.
You are going to need to know basic coding to get what you need to do, done.
What are you using elliptical integrals for - not this, surely?
Brett
-
You are going to need to know basic coding to get what you need to do, done.
What are you using elliptical integrals for - not this, surely?
Brett
I can write code Brett, but it's not my forte. I learned Fortran in college and can write C++ to a level. My skills are enough to build the basic control. I can hack others code into a package I can use. I find it refreshing actually, to learn more.
Elliptical integrals - Nothing I can avoid. I think the most extensive would by integrating the work over a spherical path and that would be simply to show proof of a concept derivable via point conditions. The 2 D derivation makes intuitive enough sense to expand to 3 space. My math skills have atrophied in the 35 plus years sense I last used them on that level. From what I have been able to discern, that is something well within your umbrella of tools.
-
Elliptical integrals - Nothing I can avoid. I think the most extensive would by integrating the work over a spherical path and that would be simply to show proof of a concept derivable via point conditions. The 2 D derivation makes intuitive enough sense to expand to 3 space. My math skills have atrophied in the 35 plus years sense I last used them on that level. From what I have been able to discern, that is something well within your umbrella of tools.
Well, maybe, but I am very curious as to the applicability to this particular problem - you aren't attempting to come up with a closed-form equation for the airplane path, are you? I certainly would not do that, and I don't think you need to, since you can numerically integrate more-or-less anything you need to, to arbitrary accuracy. Either that, or you are way ahead of us on something!
Brett
-
Well, maybe, but I am very curious as to the applicability to this particular problem - you aren't attempting to come up with a closed-form equation for the airplane path, are you? I certainly would not do that, and I don't think you need to, since you can numerically integrate more-or-less anything you need to, to arbitrary accuracy. Either that, or you are way ahead of us on something!
Brett
I may be way ahead of you on this or not, but it isn't because of integrating the path. That truly isn't necessary to design a control. That is the kind of thing we would do for an engine though but again that would be more of a 3d look up table and we interpolate between points. I only mentioned it as a mathematical proof. Not for any other reason. I need to finish writing the performance analysis and present that. I've started but, as you know, written words don't convey concept well. Illustrating takes allot of time. This effort is good enough to warrant illustration better than the pictures of my white board.
-
There is a simple reason for using C++, that just about trumps any argument against it: the entire Arduino ecosystem is built around it.
All the libraries, hardware, examples, forums, etc are a huge resource, that (as far as I know) just is not matched in any other language.
You always have the option of using assembler with Arduino when you think it is more appropriate.
You also don't need to get into the weeds of object oriented stuff, and the code can end up looking a lot like regular C code.
-
There is a simple reason for using C++, that just about trumps any argument against it: the entire Arduino ecosystem is built around it.
All the libraries, hardware, examples, forums, etc are a huge resource, that (as far as I know) just is not matched in any other language.
You always have the option of using assembler with Arduino when you think it is more appropriate.
You also don't need to get into the weeds of object oriented stuff, and the code can end up looking a lot like regular C code.
For a guy like me, this is exactly why I use it. I'm not a code guy and I can hack out just about anything I wish with these resources.
-
I'm working on an electric powerplant control / timer and I'm fairly certain that a near Constant Velocity could be achieved. I'll publish that work eventually. It has been the focus of much of the testing I have been doing with the exception of the spade exercise which was a valuable distraction. In the process of the analysis, I make the assumption (simplification) of Constant Velocity, CV, in order to simplify the math involved. We always talk about that being desirable but I'm not so sure. While flying the SV videos I would swear the velocity is nearly constant but when I do the math and watch the video for verification, I see the airplane decelerating on the 45 lines. Which begs the constant Angular Rate question.
A Constant Angular Rate would make the squares look uniform and the leg time would be the same basically. With a CAR the upper level line would take the same time as the lower level line. With CV the top leg would be short time while the bottom leg would be long time. When we watch objects moving we don't necessarily see the actual speed so much as we see the characteristic length coved per unit time. A 767 on final looks like it is flying much slower than a Learjet 23 does even though both have nearly identical approach velocities. My truck going down the road at 80 MPH feels much slower than the wife's LaCross doing 80 MPH.
So, the question opened for discussion is CV or CAR as it relates to perception of the quality of the maneuvers.
I've judged quite a few PA flights with no griping from flyers. I NEVER paid attention to the speed of the plane. It was all about "did the plane follow a steady, smooth flight path. Did the plane follow the flight path that matched the maneuver descriptions, how bad were any errors. Did they do the maneuvers in the correct order, proper number of intermediate laps, did the landing occur as the last maneuver.
There's nothing in the rules about keeping an even airspeed. If the plane could hold a constant airspeed it might make the maneuvers appear more uniform.
-
There's nothing in the rules about keeping an even airspeed. If the plane could hold a constant airspeed it might make the maneuvers appear more uniform.
I'd expect that some judges would see a deviation from the timing they're used to, feel disturbed about it (consciously or unconsciously) and you'd get lower scores. OTOH, I think that the good judges wouldn't do that, and the "bad" ones would get used to it for seeing it all the time.
-
If the plane could hold a constant airspeed it might make the maneuvers appear more uniform.
I have been following this thread intently because I really am interested in why our planes do what they do. I haven't commented so far because I don't speak engineerese. This is in English! I agree that when you are judging, or for that matter just watching, the actual airspeed doesn't really catch your attention, ground speed maybe but airspeed? I am not sure that how it appears from the outside is as important (to me at least) as how it feels from the inside. I would like consistent airspeed, except maybe in the down leg of the hourglass. The active timers attempt to emulate that but the lag in when you need it, and when you get it, can be a bit annoying BUT, you at least know it is coming. One of the more difficult times I am having adjusting to both logarithmic flaps and an active timer is the varying feel in the rounds. Having a timer that constantly kept the airspeed the same, or close, would really help me.
Ken
-
I have been following this thread intently because I really am interested in why our planes do what they do. I haven't commented so far because I don't speak engineerese. This is in English! I agree that when you are judging, or for that matter just watching, the actual airspeed doesn't really catch your attention, ground speed maybe but airspeed? I am not sure that how it appears from the outside is as important (to me at least) as how it feels from the inside. I would like consistent airspeed, except maybe in the down leg of the hourglass. The active timers attempt to emulate that but the lag in when you need it, and when you get it, can be a bit annoying BUT, you at least know it is coming. One of the more difficult times I am having adjusting to both logarithmic flaps and an active timer is the varying feel in the rounds. Having a timer that constantly kept the airspeed the same, or close, would really help me.
Ken
Well, my initial intent was not to make this an engineering discussion and more about perception. But then I went said I was working on a control which I have been for quite some time and that let all us ubergeeks out to play on the drawing CAD board table.
-
When you say constant velocity, what is that relative to? The ground, or the air?
On squares, the top leg IS shorter than the bottom leg. Don't know how you could get to the same time by either method.
Only because I've got nothing better to do today I thought I'd make a short response to the above short bit of Paul's previous post: while Paul is correct with reference to how we do things in "stunt" I wonder if we're doing the right thing.
Stunt Pattern Madness
"Snip!" "On squares, the top leg IS shorter than the bottom leg." "Snip"
Oh, dear. I'm gonna get my butt kicked...nonetheless...into the breech.
That the stunt community's absurd insistence of describing a correct pattern maneuver based on the point of view of the Judges (who might be observing said maneuver from any damn place during a flight) has been dithered about ad nauseam for decades on and about the CLPA community drives me nuts. Trying to address the stunt pilot's utilization of the flight sphere from the point of view of judges is the consummate definition of insanity.
There is only one place where the accuracy of the performance of “figures” on the surface of a sphere is accurate: in the center of the circle. The only person in a position to accurately determine the success or failure correctly is the pilot…for whom every point on the surface “is” “identical”! Round is round, square is square, a straight line a straight etc. etc.
A square loop can be accurately seen, performed and evaluated only from the one seat in the auditorium…one already taken…where the pilot is standing. Where the surface of the performance (hemi) sphere is the the same everywhere and thus the display and view of straight lines (hemispheres), perfect loops (circles) and figure eights (two adjacent perfect circles), etc. etc. etc. are viewed by the pilot.
A straight flight path…although a great circle arc on the sphere’s surface…from the pilot’s view is, in fact a straight line…etc. etc. etc. A perfect clover is four perfectly round and sized loops neatly “just” touching each other(from the pilot’s view) , etc. etc.
The pilot should be charged with doing just that: i.e. making the tricks look exactly like their titles describe from the only point they can be viewed so as to determine that perfection!
The judges are called judges because it is their responsibility to view the maneuvers and…to the best of their ability…try to mentally place themselves at the pilot’s location and, to the greatest degree possible make an educated assessment of how well the pilot has performed the prescribed maneuver.
The Pilot is paid to fly the tricks (as described by an elementary school teacher to his/her students) from his singular critical observation point. We pay (yea right!) the judges the big bucks for doing their absolute best to put themselves in the pilot’s shoes and, as best they can, assign a value to how well the pilot has done his/her job. Neither job is easy but both should be doing their best to do their best. (Notice the large number of the words “their best” are included in that description of the tasks involved)
IMHO The worst thing we’ve done in assigning the responsibilities of Pilots and judges is to try to require the pilot to distort the shapes of their tricks--viewed from the perfect location—and to do so accurately so as to be “wrong” in just the right manner to the point that the trick will look more or less OK to the judges (from wherever they might be viewing the result).
-
I'm in total agreement, Ted.
dg
-
IMHO The worst thing we’ve done in assigning the responsibilities of Pilots and judges is to try to require the pilot to distort the shapes of their tricks--viewed from the perfect location—and to do so accurately so as to be “wrong” in just the right manner to the point that the trick will look more or less OK to the judges (from wherever they might be viewing the result).
[/color]
Who *are* these people we call judges? For the most part they are *us* and we should know better. I have been told that I needed to distort maneuvers so that the judges could see them correctly. But, when I am judging do they think I don't know what they look like so they have to do them wierd just for me? I think a lot of this a hold over from the old Navy days where the judges really didn't know what they were looking at!
Ken - And I am in total agreement with Dale!
-
Who *are* these people we call judges? For the most part they are *us* and we should know better. I have been told that I needed to distort maneuvers so that the judges could see them correctly
I can guess who that might have been, however, you have to get that *right out of your head*. Even if it was a good idea - which it isn't - most people cannot even do the undistorted maneuvers well enough to be recognizable, much less put in tiny distortions that you might imagine "correct" the shapes. In the unlikely event we can reliably fly accurately to about 3" - instead of, very commonly, 10-20 feet - there's no point in worrying about putting in intentional distortions.
Brett
-
I put in the corrections. For example, when there are three judges spread out, I often put one outside loop directly across from each judge.
-
I pretty much just fly and don't worry about the judges. Draw my best figures and let it be at that.
-
... I have been told that I needed to distort maneuvers so that the judges could see them correctly ...
... In the unlikely event we can reliably fly accurately to about 3" ...
The biggest contest I've gone to has been the NW Regionals in Oregon -- there, out of a field of 15 to 20 pilots flying Expert, it's really only the top two or three who fly well enough that they hit that close. These are people who regularly win the Nationals and/or the Worlds, and it is out of a field of pilots dedicated enough that they're driving hundreds of miles to a contest that's conveniently timed to be two months before the Nationals.
So -- don't worry about it until you're getting into the top 5 at the Nats, or think you have a serious chance of winning the team trials.
-
The biggest contest I've gone to has been the NW Regionals in Oregon -- there, out of a field of 15 to 20 pilots flying Expert, it's really only the top two or three who fly well enough that they hit that close. These are people who regularly win the Nationals and/or the Worlds, and it is out of a field of pilots dedicated enough that they're driving hundreds of miles to a contest that's conveniently timed to be two months before the Nationals.
So -- don't worry about it until you're getting into the top 5 at the Nats, or think you have a serious chance of winning the team trials.
I don't worry about it at all because I don't subscribe to it as either a competitor or a judge. I can understand the controversary though. The rule book is very unclear on anything with corners. A simple statement at the beginning stating that all maneuvers are described as viewed by the flier and projected on a flat surface. Then you can properly use words like square. Anybody that has bent solder over a bowling ball knows this. I am trying to see how any of this relates to the thread subject, and I can't.
ken
-
It shouldn't be any controversy, just a conversation about perception. If it's the top 5, 10 or twenty or just a local person, it shouldn't matter. What are the aesthetics involved. My personal thinking is that a constant velocity would look odd on any level. I think up high going slower and subscribing a comparative angular rate as the lower side is potentially aesthetically more pleasing.
-
I put in the corrections. For example, when there are three judges spread out, I often put one outside loop directly across from each judge.
Howard,
Do they frequently come up to you afterward and say thanks??? ~^ ~^ n~
Ted
-
I pretty much just fly and don't worry about the judges. Draw my best figures and let it be at that.
There ya go, Mark. That's pretty much all we can do, isn't it!?
Ted
-
I don't worry about it at all because I don't subscribe to it as either a competitor or a judge. I can understand the controversary though. The rule book is very unclear on anything with corners. A simple statement at the beginning stating that all maneuvers are described as viewed by the flier and projected on a flat surface. Then you can properly use words like square. Anybody that has bent solder over a bowling ball knows this. I am trying to see how any of this relates to the thread subject, and I can't.
ken
Right on, Ken. The rule book and pattern pundits are predictably unclear as they attempt to describe the indescribable. Perhaps, even, not so much the rule book as the acres and acres of faux computer paper spent on trying to do so!
Just as a simpler for instance, I (and I'm sure many others) did pretty well for a pretty long time by flying the tricks where the air best advantaged trying to make them look right "only" to me...and trusted the judges to grade them to the best of their ability from their point of view.
Ted
p.s. If we included a vertical "square figure eight" in the pattern would the resulting rule book description require that the topmost half of the figure consist of a single corner...an angular tilted up leg reversed to a matching diving tilted down leg. Sorry, just being a smart arse.
-
p.s. If we included a vertical "square figure eight" in the pattern would the resulting rule book description require that the topmost half of the figure consist of a single corner...an angular tilted up leg reversed to a matching diving tilted down leg. Sorry, just being a smart arse.
I think it would be a cool maneuver to add. I would have to think about the wording, but it would be bounded by a rectangle that is defined by 45° worth of flight at the bottom of the circle, 45° worth of flight along the "wingover path", and straight-looking connecting paths. Those connecting paths would then have to not be great circle paths so you'd get all sorts of interesting things with the angles not being 90° blah blah blah.
I think I would make a really impressive looking and pretty maneuver. You could put it right between the vertical eights and the hourglass, and then the order of the vertical a eights would be exactly like the order of the loops.
-
On squares, the top leg IS shorter than the bottom leg. Don't know how you could get to the same time by either method.
The AMA square loops do not require that the side legs be 90o (perpendicular) to the ground, so the top leg does not necessarily need to be shorter than the bottom leg. The thing that the AMA rules require is that the top leg be at a constant 45o elevation. Otherwise, if that top leg could be a great circle path as are the bottom and side legs, the maneuver could be flown with all for sides being the same length (all great circle paths) and all four corners being the same angle (which would be slightly less than 90o on the plane tangent to the circle at each corner.) Unfortunately, the rule requires the 45o constant elevation for the top side.
Keith
-
p.s. If we included a vertical "square figure eight" in the pattern would the resulting rule book description require that the topmost half of the figure consist of a single corner...an angular tilted up leg reversed to a matching diving tilted down leg. Sorry, just being a smart arse.
That would be an interesting figure to be sure but it would not be flown on the cardinal ordinates of the sphere it would have to have some odd great circle elements to it. It definitely would be fun to explore. What about an overhead square eight? As long as we're inventing maneuvers, how about a diamond eight or a bowtie (hourglass on it's side).
-
As far a a square maneuver being the same length on all four sides, use the engineers favorite check by taking things to the limit. A square maneuver by definition has vertical sides that are perpendicular to the ground. Now, consider a 90 degree tall square loop. It is 90 wide. Follow those sides vertically to 90 degrees, and what do you get? They intersect, forming a three sided square. Oops..three sides? If the "square" is limited to 45 degrees, the top horizontal leg IS shorter than the bottom level flight leg. If you use timing, to do your square, it will not be correct. Gravity against you going up, shorter leg on the top, gravity with you going down, and the bottom leg longer than the top. All contribute so one shouldn't use your internal metronome for this maneuver.
As has been mentioned before, the AMA rules do not require that the sides be perpendicular to the ground. (The FAI rules do require the sides be vertical to the ground which results in the top side being noticeably shorter than the bottom leg.) So, the AMA rules allows the top leg approach the same length as the bottom leg, the side legs need not be perpendicular to the ground. The bottom corners are not required to be 90o to the ground. The AMA square can be flown with all sides being equal length, are all great circle paths and the corners are slight less than 90o. I maintain that much a maneuver will look very nice to the judges as the timing will appear/sound more consistent and the dip in the top leg due to parallax (from the judges perspective) will much less noticeable.
You have described a perfect triangle traced on our hemisphere where all three sides are equal length (great circle paths), the upright sides are vertical to the ground, and all three corners are 90o.
It is possible to construct an equangular triangle of any size up to where the top is at the 90o point above the center of the circle on our hemisphere where all three sides are the same length (great circle paths) and all three corners are the same. This can be of any size from infinitesimally small with three 120o corners to the one just described with all three side being equal with three 90o corners. Interesting exercise because this shows that the AMA triangles have all three sides being equal length (great circle paths) and the corners will be something less that 120o. It would be interesting to see the math involved to actually calculate what the angle is for these three turns on our hemisphere for the AMA triangle that has the top at the 45o elevation. These turns are definitely less than 120o.
I would like to find a good test book on the basics of spherical geometry.
Keith
-
It is possible to construct an equangular triangle of any size up to where the top is at the 90o point above the center of the circle on our hemisphere where all three sides are the same length (great circle paths) and all three corners are the same. T
I would like to find a good test book on the basics of spherical geometry.
Keith
The answer is yes, an equilateral triangle is not only possible but the fundamental derivation the trigonometry on the sphere. The length of a side is the radius time the angle in radians. So three side of equal length by definition have three equal angles.
Fortunately my recent endeavors have caused me to brush off some very old math cobwebs in spherical geometry. There's lots of good references in the world. Duck Duck Go it for Spherical math or trigonometry.
-
As has been mentioned before, the AMA rules do not require that the sides be perpendicular to the ground. (The FAI rules do require the sides be vertical to the ground which results in the top side being noticeably shorter than the bottom leg.) So, the AMA rules allows the top leg approach the same length as the bottom leg, the side legs need not be perpendicular to the ground. The bottom corners are not required to be 90o to the ground. The AMA square can be flown with all sides being equal length, are all great circle paths and the corners are slight less than 90o. I maintain that much a maneuver will look very nice to the judges as the timing will appear/sound more consistent and the dip in the top leg due to parallax (from the judges perspective) will much less noticeable.
You have described a perfect triangle traced on our hemisphere where all three sides are equal length (great circle paths), the upright sides are vertical to the ground, and all three corners are 90o.
It is possible to construct an equangular triangle of any size up to where the top is at the 90o point above the center of the circle on our hemisphere where all three sides are the same length (great circle paths) and all three corners are the same. This can be of any size from infinitesimally small with three 120o corners to the one just described with all three side being equal with three 90o corners. Interesting exercise because this shows that the AMA triangles have all three sides being equal length (great circle paths) and the corners will be something less that 120o. It would be interesting to see the math involved to actually calculate what the angle is for these three turns on our hemisphere for the AMA triangle that has the top at the 45o elevation. These turns are definitely less than 120o.
I would like to find a good test book on the basics of spherical geometry.
Keith
Check out AMA CLPA rules, section 13.10. HS8..
It says vertical sides are 90 degrees from level, and the top segment IS shorter than the other legs.
Sounds like more work on the rules is required....
-
Check out AMA CLPA rules, section 13.10. HS8..
It says vertical sides are 90 degrees from level, and the top segment IS shorter than the other legs.
Sounds like more work on the rules is required....
I thought they made that change a few years ago, to match up with the FAI definition.
-
Check out AMA CLPA rules, section 13.10. HS8..
It says vertical sides are 90 degrees from level, and the top segment IS shorter than the other legs.
Sounds like more work on the rules is required....
Interesting. 13.10 describes the square eights. The turns/vertical paths for the square eights are defined differently than the turns/vertical paths of the square loops 13.6 and 13.7. Again, the description for the square loops do not require the sides of the maneuver to be perpendicular to the ground which allows the maneuver to be flown such that the tops and bottoms can be equal or nearly equal
Your friendly Control Line Aerobatics Contest Board would certainly entertain suggestions/proposals for improvements. A new two-year rules change cycle has just started.
It seems that every time I read through the rule book, I find things that could be improved. It only took nearly 20 years to get the four leaf clover correctly defined from the time when the description was changed from the start of the maneuver to be at the 38o elevation to a 42o elevation which was still incorrect. The new rule says nothing about a constant 42o elevation anywhere in the maneuver. That 38o start point goes back to the early 60's. (Some people still seem to fly it that way.)
Keith
-
I thought they made that change a few years ago, to match up with the FAI definition.
Actually, I think the AMA Control Line Aerobatics Contest Board deliberately did not change the rules for the inside and outside squares to be like the FAI rules. This allows the AMA squares to act more like square loops than the trapezoids defined by the FAI rules with their truncated top legs being noticeably shorter than the bottom legs. Having attended several World Championships, I do not think the top FAI pilots do that anyway. (My personal observations, other opinions may differ.)
Keith
-
Actually, I think the AMA Control Line Aerobatics Contest Board deliberately did not change the rules for the inside and outside squares to be like the FAI rules.
Keith
Personally, I do not like the FAI definitions for the squares either, but they are headed towards a full "great circle" definition which makes sense given that we do fly on the surface of a hemisphere. Nothing describes an event better than a picture. What would be wrong with rewriting the entire maneuvers section using accurate drawings on a half globe other than finding someone with the software and nothing better to do. The rules section could show the maneuver from the pilots' position and the judges guide show it from the judges' position. We have the technology to do that, and it is *not* the English language. It would save a lot of conflicts. For example, we do not specify that the top and bottom corners be the same on the squares. That definition allows the square to be flown but in the same definition the word square is used which, by definition, has equal radius corners and equal sides, again making the maneuver impossible to fly on a hemisphere. Plus we are left assuming that you don't consider the top and bottom as sides, which they actually are.
It is *tongue in cheek* but by using accurate drawings, the definition of a maneuver could be as simple as "Do this (See Pilots' Picture)"
The Judges guide - "Didn't Do that (See judges' picture)
We should not be performing to make it easy on judges, yet that does seem the logical road to success. IMHO, the answer is to use technology to draw accurate pictures of our maneuvers and learn to recognize a properly flown maneuver from EITHER position.
Do I think that the current crop of leadership would support such a radical change - LL~ LL~ LL~
Ken
-
For example, we do not specify that the top and bottom corners be the same on the squares. That definition allows the square to be flown but in the same definition the word square is used which, by definition, has equal radius corners and equal sides, again making the maneuver impossible to fly on a hemisphere.
Ken
I think it has long been understood that the rules require that the figures flown defined as square are actually "sort of square" or "squarish" though not explained that way. As soon as the rule says that the top leg is to be flown at a 45o constant elevation, the figure cannot have all four legs to be the same. That requirement also does not allow the "top and bottom corners be the same" which is NOT required. At least, the rule states that "The two (2) bottom corners are equal and so are the two (2) top corners." The rules for all of the square maneuvers as well as the triangles and the hour glass does specify that that the corners "shall be of a tight radius" which clearly (and thankfully) does not specify either the angular change or a specific "radius" of the turns for any of these maneuvers.
By contrast, the triangles and the hourglass can be flown with equal turns and equal legs, all of which are great circle paths.
Keith
-
IMHO, the answer is to use technology to draw accurate pictures of our maneuvers and learn to recognize a properly flown maneuver from EITHER position.
Ken
There are simulations that show in 3-D how each of the maneuvers appear on a hemisphere when flown according to the rules. (The work on this by Keith Renecle from South Africa being notable and probably the first to do so which also eventually led to the rule change that redefines the four leaf clover.) Keith's simulations show how the complete pattern appears on the hemisphere from any perspective outside of the hemisphere including where a the judge's eyeballs would be positioned. His graphics clearly shows the hemisphere and how the maneuvers described by the rule should appear. It even shows how the five foot radius would appear on those maneuvers that now only require a "sharp turn".
It is just that no one has taken the steps to develop the drawings that would be suitable to put in the rule book. This would show the hemisphere and the path for each maneuver on that hemisphere if someone would want to take on that task, I am sure the Contest Board would favorably accept that work to change the rule book.
I would think that a set of drawings from the judge's perspective is all that is necessary. To represent each maneuver from the pilot's perspective in a meaningful manner would be a bit more difficult and not necessary. It still remains that the goal of the pilot is to fly each maneuver so that it appears correct to the judges.
Keith
-
There are simulations that show in 3-D how each of the maneuvers appear on a hemisphere when flown according to the rules. (The work on this by Keith Renecle from South Africa being notable and probably the first to do so which also eventually led to the rule change that redefines the four leaf clover.) Keith's simulations show how the complete pattern appears on the hemisphere from any perspective outside of the hemisphere including where a the judge's eyeballs would be positioned. His graphics clearly shows the hemisphere and how the maneuvers described by the rule should appear. It even shows how the five foot radius would appear on those maneuvers that now only require a "sharp turn".
It is just that no one has taken the steps to develop the drawings that would be suitable to put in the rule book. This would show the hemisphere and the path for each maneuver on that hemisphere if someone would want to take on that task, I am sure the Contest Board would favorably accept that work to change the rule book.
I would think that a set of drawings from the judge's perspective is all that is necessary. To represent each maneuver from the pilot's perspective in a meaningful manner would be a bit more difficult and not necessary. It still remains that the goal of the pilot is to fly each maneuver so that it appears correct to the judges.
Keith
With our modern graphics tools that is not impossible to do. Might even be helpful. If it weren't a bucket load of work when I'm already so diverse in my efforts, I'd do it.
-
With our modern graphics tools that is not impossible to do. Might even be helpful. If it weren't a bucket load of work when I'm already so diverse in my efforts, I'd do it.
Here is what the same hourglass looks like from the perspective of the pilot and the judge. These are crude examples but they do illustrate the point. I am not sure that it is a really difficult task to draw both. I think that Keith R.'s latest version has this built in. I can remember playing with it before our fire. The version I have now is quite old. Problem is that it is programmed for the FAI version not the AMA.
Notice that the picture taken from the inside makes the figure look like it was drawn on a flat surface. The one from outside shows the curve of the hemisphere. That is what a judge sees.
Ken
-
With our modern graphics tools that is not impossible to do. Might even be helpful. If it weren't a bucket load of work when I'm already so diverse in my efforts, I'd do it.
There are only about 14 months left to make a formal proposal.
Keith
-
There are only about 14 months left to make a formal proposal.
Keith
Oh I don't have any intent on making a rules proposal. My input is only a potential graphics contribution. I'm not so serious about competing such that I wish to start making rules proposals on figures. If I were to get serious about competing and bringing any proposals it would be much more extensive than simply making "better" different figure graphics. In the 4/4 arena we use Aresti figures and they are even less well exact in their diagrams. They are a short hand language on their own. I've tested that water a bit and it isn't anything I think would be well received.
I'll attach link to a condensed version and a snippet for comparison.
http://www.iac52.org/2012/Catalog2012ChangedPagesFinal.pdf
-
I have a friend who does a single entendre. I didn’t see a single entendre in the catalog. Is that a humpty bump variant?
-
I have a friend who does a single entendre. I didn’t see a single entendre in the catalog. Is that a humpty bump variant?
I know, crazy right? What's entertaining is to fly an unknown sequence made out the catalog figures without ever having flown it before.
-
That is doable. A variable pitch propeller is necessary to achieve it.
Looks like I missed something here, but since question was answered in 3rd comment and I am not going to mess with C++ problems and sphere geometry, I will put only few comments.
So - As Paul mentioned, the light prop was developed especially for easier RPM controll. As Brett mentioned its (the prop) and motor mass inertia moment limits bandwidth, it is very imprtant for controll. That is also reason why I stick with small diameter motors (2826 - like AXI). We tested also MVVS which has stator diameter 35mm and difference was obviouse.
BUT there is also another problem, thrust controlled by RPM is asymetric, means prop can better pull than brake. Reasons are 2 - the first is chambered airfoil, that was another object of optimization for new props, and second is lower RPM during braking also means lower reaction to model speed changes (the same prop at higher RPM will change the thrust as function of speed difference more that the same prop at lower RPM).
That means that pitch controlled prop will not only react quicker (with quick servo - I do not think it will be so quick with normal servo - reaction time of classic servos is not much higher that reaction of my power train, I cen easily show that frequency of of RPM change input can be higher than standar R/C servo - we will use probably small light servo which can make run from one side to the other loaded at say 0.2s what is visible lag; but high speed servos working at 400Hz inputs will be probably enough).
However I am not sure if I want have such prop on contest model. Fixed prop is fixed prop and I am sure it will never dismuont during official flight, I can be sure that I can easily replace if we fly on typical WCh potato fields etc ;D
But who knows, may be Mark will push it to reliable solution then I will think about it.
-
However I am not sure if I want have such prop on contest model. Fixed prop is fixed prop and I am sure it will never dismuont during official flight, I can be sure that I can easily replace if we fly on typical WCh potato fields etc ;D
But who knows, may be Mark will push it to reliable solution then I will think about it.
I fully agree with you on the motor and fixed pitch braking part. That is the whole reason I launched on the variable pitch propeller effort in the first place. The reliability part is why I put a hinge in the propeller hub in the rotation plane to reduce / limit the bending fatigue. We'll see where that goes.
-
Igor, what is the moment of inertia of your 12” flat-back propeller?
-
That means that pitch controlled prop will not only react quicker (with quick servo - I do not think it will be so quick with normal servo - reaction time of classic servos is not much higher that reaction of my power train, I cen easily show that frequency of of RPM change input can be higher than standar R/C servo - we will use probably small light servo which can make run from one side to the other loaded at say 0.2s what is visible lag; but high speed servos working at 400Hz inputs will be probably enough).
However I am not sure if I want have such prop on contest model. Fixed prop is fixed prop and I am sure it will never dismuont during official flight, I can be sure that I can easily replace if we fly on typical WCh potato fields etc ;D
I would not think it out of the question that we might have to make our own servo, and it may not look like a typical RC servo. But I did a bit of scribbling and convinced myself that you can manage it with a very fast conventional servo, because throw of any reasonable pitch-changing mechanism will be tiny compared to a typical RC output arm, so at least you won't be rate-limited any more. Getting the slop out of it, or what such a fast system will do with slop/hysteresis, that's another story.
As you note, the thing that has put me off the variable-pitch system is the potential for winding up with a blade sticking out of my forehead (or someone else's forehead). Paul working on it gives me some hope, because he is a professional at this sort of problem.
And, as always, having an "ideal" system, while a neat idea, is probably not necessary. The existing systems work well enough that they are no longer the determining factor, so maybe it doesn't need to be any better. Simply defining what it is we are trying to do in clearly-stated engineering parameters seems to have eluded us to date.
One thing I do know, we aren't going to figure it out entirely as a paper exercise - while that might be very helpful and useful, at some point we are going to have to build something and see.
Brett
-
I would not think it out of the question that we might have to make our own servo, and it may not look like a typical RC servo. But I did a bit of scribbling and convinced myself that you can manage it with a very fast conventional servo, because throw of any reasonable pitch-changing mechanism will be tiny compared to a typical RC output arm, so at least you won't be rate-limited any more. Getting the slop out of it, or what such a fast system will do with slop/hysteresis, that's another story.
As you note, the thing that has put me off the variable-pitch system is the potential for winding up with a blade sticking out of my forehead (or someone else's forehead). Paul working on it gives me some hope, because he is a professional at this sort of problem.
And, as always, having an "ideal" system, while a neat idea, is probably not necessary. The existing systems work well enough that they are no longer the determining factor, so maybe it doesn't need to be any better. Simply defining what it is we are trying to do in clearly-stated engineering parameters seems to have eluded us to date.
One thing I do know, we aren't going to figure it out entirely as a paper exercise - while that might be very helpful and useful, at some point we are going to have to build something and see.
Brett
I wouldn't use a servo for more than just the positioning aspect reason. Directly driven the pitch mechanism has to carry the propeller hinge moment continuously and a servo would have to carry that. This would lead to high current drain and potential oscillation. Blades separating is a very real concern, catastrophic to the airplane and hazardous to people. My propeller design is past mere paper and is close to a stage it can be sent to a machine. I'm honestly holding for more test time on the hinge propeller before proceeding. Part of that test time involves the flight data package I'm working on.
-
Igor, what is the moment of inertia of your 12” flat-back propeller?
If I remember well, it was somewhere at 70 to 75 ukg.m^2. But I am sure it was less than 11" prop as it has wider blades (it is more than 1g heavier). And aproximately 2/3 or full carbon prop from the same mold.
-
I would not think it out of the question that we might have to make our own servo, and it may not look like a typical RC servo. But I did a bit of scribbling and convinced myself that you can manage it with a very fast conventional servo, because throw of any reasonable pitch-changing mechanism will be tiny compared to a typical RC output arm, so at least you won't be rate-limited any more. Getting the slop out of it, or what such a fast system will do with slop/hysteresis, that's another story.
...
One thing I do know, we aren't going to figure it out entirely as a paper exercise - while that might be very helpful and useful, at some point we are going to have to build something and see.
There is one problem with them, they have dead band, so if you want some kind of precision, and AoA range for our application is narrow, then it is necessary to make some gear/arm ratio that servo runs in its full range anyway and we are back at that speed problem. However as I said before there are high speed digital types with little narrower dead band (but still present), but beside price there is weight and size penalty.
It is solvable problem, I just wanted to say it is not so trivial and "automatical" as it looks for first look.
But I can only agree with last sentence, it looks the solution is easy until you try it in real VD~ ... then come those "yes but"s and that simple solution needs bugfixes and improvements until you find that it was completaly wrong idea and with gained experience finally come usefull solution ... 3 or 4 years later ;D
-
There is one problem with them, they have dead band, so if you want some kind of precision, and AoA range for our application is narrow, then it is necessary to make some gear/arm ratio that servo runs in its full range anyway and we are back at that speed problem. However as I said before there are high speed digital types with little narrower dead band (but still present), but beside price there is weight and size penalty.
It is solvable problem, I just wanted to say it is not so trivial and "automatical" as it looks for first look.
But I can only agree with last sentence, it looks the solution is easy until you try it in real VD~ ... then come those "yes but"s and that simple solution needs bugfixes and improvements until you find that it was completaly wrong idea and with gained experience finally come usefull solution ... 3 or 4 years later ;D
Isn't that the truth. I'm at least a year in to my variable pitch efforts and am still working on getting good data. Of course I've spent some time working on airplanes too which hasn't helped the effort.